1 On tough-movement* Milan Rezac, University ... - Multimania.co.uk

. This book is difficult to convince people/anyone that they ought to read e. (Chomsky
1981:314 as "more or less acceptable" though less than an infinitival example).
c. ?This boulder would be easy for me to claim that I had lifted e. (Heycock 1994:260)
d. Mary is tough for me to believe that John would ever marry e. (Kaplan and Bresnan
1982)
e. %Mary is hard for me to believe Leslie kissed e. (Dalrymple and King 2000:16)
f. That kind of mistake is hard to realize you're making e. (Calcagno 1999)
Parasitic gaps are licensed by OP-gap, (49) (Chomsky 1982, 1986ab). I will not use parasitic
gaps further, as their analysis continues to mysteries (see Browning 1989: chapter 3, Cinque
1990: chapter 3, Nissenbaum 2001, and the articles in Culicover and Postal 2001).
(49) Cernunnos i was probably not easy [OP i to poke e i [without offending pg i ]].
The OP clause is a barrier to further extraction out of it, as expected if headed by the top of
an Ā-chain. There are superficial counter-examples such as (50)a, which all involve D-linked
wh-phrases whose gap must moreover be situated in a right-peripheral position (Chomsky
1981:311) and must not be embedded deeper within the OP clause (Jacobson 2000). Otherwise,
extraction from the OP clause is impossible, (51).
(50) a. Which violins 2 are the sonatas 1 easy to play e 1 on e 2 ? (Chomsky 1981:310)
b. *Which violin 2 is that sonata 1 hard to imagine (anyone) playing e 1 on e 2 / wanting to
play e 1 on e 2 ? (Jacobson 2000)
(51) a. *How intelligent 2 is John 1 easy to think of / regard e 1 as e 2 ?
b. How intelligent 1 is it easy to think of / regard John as t 1
c. ?John 1 is easy to think of / regard e 1 as very intelligent.
(cf. Chomsky 1981:311)
A further constraint is that the TM OP-gap chain must be contained within any wh-gap
chains, obeying Pesetsky's (1982) Path Containment Condition on interacting Ā-dependencies:
(52) a. *Which sonatas 2 are the violins 1 easy to play e 2 on e 1 ? (Chomsky 1981:310)
b. *Which people 2 are the books 1 easy to convince e 2 to read e 1 ? (Chomsky 1981:310)
This degradation is very sharp, sharper than in corresponding crossing overt wh-movement
chains. There is quite a simple explanation available within theories that start out from Rizzi's
(1990) relativized minimality: if the OP-gap and the wh-gap are both equally attractable by the
Ā-probe at the top of the OP clause, only the closer gap can be found. The latter fact may be part
of a larger pattern; Rizzi (forthcoming) observes that different types of Ā-movement such as wh
and focus movement are frozen for each other, which would fall out if there was a unitary Ā-
probe, with criterial positions differentiated by the interpretable features of the target (op.cit.,
note 2). If the gap closest to C OP is the one eventually λ-bound by the TM subject, all is well, and
the data-set in (50) results. If it is an overt wh-phrase, the derivation does not converge because a
number of mismatches arises: the wh-phrase will not have its [wh] feature deleted, it cannot
move to the top of the OP clause which does not tolerate overt material (4.4), the matrix C will
18